RU2657173C2 - Sentiment analysis at the level of aspects using methods of machine learning - Google Patents

Abstract

FIELD: physics.

SUBSTANCE: they perform a syntactic-semantic analysis of a part of the text in the natural language for obtaining a set of syntactic-semantic structures. They perform an interpretation of syntactic-semantic structures using a set of production rules for detecting in the part of the text in the natural language an aspectual term representing the aspect associated with the target entity. The value of the classifier function is calculated, using the text characteristics obtained in the syntactic-semantic analysis, for determining the key note associated with the aspectual term. They create a report that contains a hierarchical list of aspectual terms that include the identified aspects and key tones of the identified aspects. The parameter of the classifier function is determined using a training data sample and a confirming data sample. The training data sample includes educational text in natural language containing a variety of aspectual terms.

EFFECT: improving the accuracy of sentiment analysis of texts in natural language.

14 cl, 21 dwg

Description

FIELD OF THE INVENTION

[0001] The present invention generally relates to the extraction of information from texts in natural languages using computer systems, and more specifically to sentimental analysis at the aspect level using machine learning methods.

BACKGROUND

[0002] Sentiment analysis is an important operation in the automatic processing of texts in natural languages and has many uses for solving real problems. Sentiment analysis at the aspect level (ABCA) can be used to automatically identify aspects of the specified target entities and sentiments expressed in relation to each aspect in natural language texts.

SUMMARY OF THE INVENTION

[0003] In accordance with one or more aspects of the present invention, the method of sentimental analysis at the aspect level described in the example may comprise: performing, by means of a computer system, semantic-syntactic analysis of at least a portion of the text in natural language to create a plurality of syntactic-semantic structures representing part of the text in natural language; interpretation of syntactic-semantic structures with the help of a set of production rules for detecting in a part of a text in a natural language at least one aspect term representing an aspect related to the target entity; and evaluating, using one or more characteristics of the text obtained by syntactic-semantic analysis, a classifier function to determine the tonality associated with an aspect term.

[0004] In accordance with one or more aspects of the present invention, the system for sentimental analysis at the aspect level described in the example may include: a memory and a processor operating with this memory, the processor being configured to: perform semantic-syntactic analysis using a computer system at least parts of the text in natural language to create many syntactic-semantic structures that represent part of the text in natural language; interpret syntactic-semantic structures with the help of a set of production rules for detecting at least one aspect term representing a aspect related to the target entity in a part of the text in the natural language; and evaluate using one or more characteristics of the text obtained by syntactic-semantic analysis, a classifier function to determine the tonality associated with an aspect term.

[0005] In accordance with one or more aspects of the present invention, an example of a permanent computer-readable storage medium may include executable instructions that, when executed by a computing system, result in operations performed by the computing system including: performing at least semantic-syntactic analysis on the computing system at least part of the text in natural language to create many syntactic-semantic structures that represent part of the text in natural language; interpretation of syntactic-semantic structures with the help of a set of production rules for detecting in a part of a text in a natural language at least one aspect term representing an aspect related to the target entity; and evaluating, using one or more characteristics of the text obtained by syntactic-semantic analysis, a classifier function to determine the tonality associated with an aspect term.

BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The present invention is illustrated by way of examples, and not by way of limitation, and can be better understood by considering the following description of preferred embodiments in conjunction with the drawings, in which:

[0007] In FIG. 1 is a flowchart of an aspect-level sentiment analysis method described in an example in accordance with one or more aspects of the present invention;

[0008] In FIG. 2 shows an example of a natural language text for use in an example of a sentimental analysis method at the aspect level in accordance with one or more aspects of the present invention;

[0009] In FIG. 3 illustrates an example of a semantic hierarchy used in the sentiment analysis method at the aspect level described in the example in accordance with one or more aspects of the present invention;

[00010] In FIG. 4 schematically shows an example of a method for interpreting syntactic-semantic structures using a set of production rules in accordance with one or more aspects of the present invention;

[00011] In FIG. 5 shows examples of production rules used in an example of a sentiment analysis method at the aspect level for detecting aspect terms in accordance with one or more aspects of the present invention;

[00012] In FIG. 6 shows examples of production rules used in an example of a sentimental analysis method at the aspect level to determine the tonality of an aspect term in accordance with one or more aspects of the present invention;

[00013] In FIG. 7 is a flowchart of an example of a method 700 for interactive sentiment analysis at the aspect level in natural language text in accordance with one or more aspects of the present invention.

[00014] In FIG. 8 shows an example of a natural language text for use in a sentimental analysis method at the aspect level in accordance with one or more aspects of the present invention;

[00015] In FIG. 9 shows the results of the text markup example shown in FIG. 8 in accordance with one or more aspects of the present invention;

[00016] In FIG. 10 is a flowchart of an example method for performing semantic-syntactic analysis of a sentence in a natural language in accordance with one or more aspects of the present invention;

[00017] In FIG. 11 schematically shows an example of the lexical-morphological structure of a sentence in accordance with one or more aspects of the present invention;

[00018] In FIG. 12 schematically shows language descriptions representing a natural language model in accordance with one or more aspects of the present invention;

[00019] In FIG. 13 schematically shows examples of morphological descriptions in accordance with one or more aspects of the present invention;

[00020] FIG. 14 schematically shows examples of syntactic descriptions in accordance with one or more aspects of the present invention;

[00021] In FIG. 15 schematically shows examples of semantic descriptions in accordance with one or more aspects of the present invention;

[00022] FIG. 16 schematically shows examples of lexical descriptions in accordance with one or more aspects of the present invention;

[00023] In FIG. 17 schematically shows examples of data structures that can be used in one or more methods implemented in accordance with one or more aspects of the present invention;

[00024] FIG. 18 schematically shows an example of a graph of generalized components in accordance with one or more aspects of the present invention;

[00025] In FIG. 19 shows an example of a syntax structure corresponding to the sentence of FIG. eighteen;

[00026] In FIG. 20 shows a semantic structure corresponding to the syntax structure in FIG. 19;

[00027] In FIG. 21 is a diagram of an example computer system implementing the methods of the present invention.

DESCRIPTION OF PREFERRED EMBODIMENTS

[00028] This paper describes methods and systems of sentimental analysis at the level of aspects. The systems and methods described herein can be used in a variety of natural language processing applications, including machine translation, semantic indexing, semantic search (including multilingual semantic search), document classification, electronic research, etc.

[00029] As used herein, the term "computing system" means a data processing device equipped with a universal processor, memory, and at least one communication interface. Examples of computing systems that can use the methods described in this document are, in particular, desktop computers, laptops, tablet computers and smartphones.

[00030] The word "Sentiment" in this document means a verbal expression of the author’s attitude or emotions towards a certain entity (for example, an object of the real world or a class of objects of the real world). A certain text in a natural language can express the author's sentiments for various aspects of one or more entities. The word “Aspect” in this document means a specific property of the entity described in the text in a natural language. In various illustrative examples, an aspect may define a characteristic, component, or function of a corresponding entity. The phrase “Aspect term” in this document means one or more words in a natural language related to an aspect of a particular entity. Sentiment can be characterized by the tonality associated with the corresponding aspect term, for example, positive, neutral or negative.

[00031] In an illustrative example, a consumer goods producer or service provider may be interested in measuring consumer sentiment for a particular product by analyzing customer feedback online, which may be in the form of unstructured or loosely structured texts. In accordance with one or more aspects of the present invention, a computing system may analyze natural language text representing consumer reviews to identify aspect terms related to one or more target entities (e.g., goods or services), and further find sentiment expressed by consumers regarding each of these aspects.

[00032] In some embodiments, a computing system that implements the methods described herein may perform two-stage sentimental analysis at the aspect level, during which the first stage defines the aspect terms and determines the tonality in the second stage.

[00033] In the first step of the process, natural language text can be analyzed, which can contain many user reviews of one or more target entities (eg, consumer goods or services). This analysis may include a set of production rules for interpreting many syntactic-semantic structures obtained as a result of syntactic-semantic analysis of text in a natural language. Production rules can produce many data objects representing the identified aspects of the target entities, which are described in more detail later in this document.

[00034] In the second step of the process, the tonality associated with each identified aspect term is determined. In some implementations of the invention, the second step may include calculating one or more classifier functions that return the tonality associated with this aspect term. The classifier function can implement various machine learning methods, such as linear classifiers, classifiers based on the tree of direct descendants, classifiers based on a random forest, classifiers based on conditional random fields (CRF), classifiers based on latent Dirichlet placement (LDA), classifiers on based on the support vector method (SVM) and / or classifiers based on neural networks (for example, World2Ves), as described in more detail later in this document.

[00035] Additionally or alternatively, tonality associated with a given aspect term can be determined by applying a set of production rules that interpret the many semantic-syntactic structures obtained by syntactic-semantic analysis of natural language text, as described in more detail below in this document.

[00036] Various aspects of the above methods and systems are described in detail later in this document by way of examples, and not by way of limitation.

[00037] In FIG. 1 is a flowchart of an example of an aspect-level interactive sentiment analysis method 100 in a natural language text in accordance with one or more aspects of the present invention. The method 100 and (or) each of its individual functions, procedures, subprograms or operations can be implemented using one or more processors of a computing system (for example, the computing system 100 in Fig. 1) in which this method is implemented. In some embodiments, method 100 may be executed in a single processing stream. In an alternative approach, method 100 may be implemented using two or more processing threads, with one or more separate functions, procedures, routines, or operations of this method being implemented in each thread. In an illustrative example, processing threads that implement method 100 can be synchronized (for example, using semaphores, critical sections, and / or other thread synchronization mechanisms). In an alternative approach, processing streams that implement method 100 can be performed asynchronously with respect to each other. Thus, although FIG. 1 and the corresponding description contain a list of operations for method 100 in a specific order, in various embodiments of the method at least some of the described operations can be performed in parallel and (or) in a random order.

[00038] In block 110, a computing system that implements method 100 can perform syntactic-semantic analysis of the source language 101 in natural language, which can be represented, for example, by one or more customer reviews of one or more target entities (eg, consumer goods or services). In FIG. 2 is an example of a natural language text 211 representing a consumer’s review of a smartphone. In various illustrative examples, a consumer review does not necessarily include the identifier of the target entity (for example, the brand and model of a particular device), however, the corresponding relationship can be deduced from the metadata associated with the consumer’s recall (for example, the recall may be associated with a category with a title, which defines the target entity).

[00039] In various illustrative examples, text in natural language can be represented by unstructured (fluent) or poorly structured text (for example, text divided into several sections defined by subheadings, such as “Product Features”, “Benefits”, “Disadvantages” etc.).

[00040] Syntactic-semantic analysis can give at least one language-independent syntactic-semantic structure 102 for each sentence in a natural language text, as described in more detail below with reference to FIG. 5-15. In some embodiments of the invention, syntactic-semantic analysis may use a user dictionary, which may contain a list of tokens that can be used in the analyzed text in natural language to refer to target entities (for example, consumer goods or services). In an illustrative example, a user dictionary may contain a list of car brands, including possible conversational variations, such as a Chevrolet and Chevy.

[00041] In some embodiments of the invention, a user dictionary may be represented by a semantic hierarchy. As schematically shown in FIG. 3, an example of a semantic hierarchy 310 may include definitions of a plurality of semantic classes, such that each class corresponds to a domain concept. Each class definition may include definitions of one or more objects assigned to a given class, including language-specific lexical classes.

[00042] To facilitate the identification of aspectual terms in the analyzed natural language text, the semantic hierarchy that is used in the syntactic-semantic analysis of natural language text can be supplemented by associating some lexical classes 315 with custom pragmatic classes 317, indicating that tokens represented by instances of these lexical classes may be used as aspect terms indicating a particular aspect. In some embodiments of the invention, a custom pragmatic class may have several subclasses to further indicate the type of aspect with which instances of the corresponding lexical class are associated. In various illustrative examples, subclasses of a custom pragmatic class may indicate that the token represented by an instance of the corresponding lexical class represents a characteristic, component, or function of the corresponding entity. In another illustrative example, a subclass of a user-defined pragmatic class may indicate that the token represented by an instance of the corresponding lexical class represents an attribute, component, or function of the corresponding entity implicitly, and tonality may be implied by the aspect term itself (for example, “high speed (RPM)” ) Otherwise, if tonality is not defined by an aspect term and can only be determined by analyzing the context of the aspect term, the aspect term may be called “explicit”.

[00043] At block 120, the computing system may interpret a plurality of syntactic-semantic structures using a set of production rules to determine particular aspects of the target entities, as described in detail below with reference to FIG. 4-5.

[00044] At block 130, the computing system may determine the tonality of the identified aspect terms. In an illustrative example, the key may be negative (-1), neutral (0), or positive (+1). In various other illustrative examples, other numerical scales and / or verbal notation of keys may be used.

[00045] In some embodiments of the invention, tonality can be determined by using a set of production rules to interpret a variety of syntactic-semantic structures resulting from syntactic-semantic analysis of natural language text. In FIG. 6 provides examples of production rules that can be used to determine the tonality of an aspect term in accordance with one or more aspects of the present description. Additionally or alternatively, tonality can be determined by calculating one or more classifier functions that return the tonality of each identified aspect term, as described in more detail later in this document.

[00046] At block 140, the computing system may aggregate the extracted information in a form suitable for display in a graphical user interface (GUI) and / or for printing. In an illustrative example, a report created by the system may contain the identifier of the target entity (for example, an electronic device) and one or more hierarchical lists identifying the components, characteristics and (or) functions of the target entity. Each hierarchical list can contain many aspect terms and the corresponding key. In some embodiments of the invention, each hierarchical list may also contain the number of instances (for example, the number of reviews on the network) in which the identified aspect is characterized by the identified aspect term and the identified tonality. In response to the completion of the operations described with reference to block 140, the method may be completed.

[00047] FIG. 4 schematically shows an example of a method for interpreting semantic structures using a set of production rules in accordance with one or more aspects of the present invention. The method 400 and / or each of its individual functions, procedures, subprograms, or operations can be implemented using one or more processors of a computing system (eg, computing system 400 in FIG. 4) that implements this method. In some embodiments, method 400 may be executed in a single processing stream. In an alternative approach, method 400 may be implemented using two or more processing threads, with each thread performing one or more separate functions, standard programs, routines, or operations of the method. In an illustrative example, processing streams implementing method 400 can be synchronized (for example, using semaphores, critical sections, and / or other thread synchronization mechanisms). Alternatively, the processing threads implementing the method 400 may be performed asynchronously with respect to each other. Therefore, although FIG. 4 and the corresponding description contain the operations of the method 400 in a specific order, various implementations of the method can perform at least some of the above operations, in parallel and (or) in an arbitrarily selected order.

[00048] In some implementations of the invention, data objects representing the identified aspects of the target entities can be represented by a graph RDF graph (Resource Definition Framework). RDF assigns a unique identifier to each information object and stores information about such an object in the form of triplets of SRO, where S stands for “subject” and contains the identifier of an object, P stands for “predicate” and defines some property of this object, and O stands for “object” and stores the value this property of this object. This value can be either a primitive data type (string, number, Boolean value), or the identifier of another object. In one illustrative example, the triplet SRO can associate a token from a text in a natural language with an aspect of the target entity.

[00049] Production rules that are used to interpret syntactic-semantic structures may include interpretation rules and identification rules. An interpretation rule may contain the left side represented by a set of logical expressions defined in one or more templates of syntactic-semantic structures and the right side represented by one or more statements related to information objects representing entities that refer to text in a natural language.

[00050] The syntactic-semantic structure template may contain certain elements of the syntactic-semantic structure (for example, checking for the presence of a specific gramme or semantema, belonging to certain lexical / semantic classes, being in a certain surface or deep position, etc.). Relations between elements of syntactic-semantic structures can be specified using one or more logical expressions (conjunction, disjunction, and negation) and (or) operations describing the relative positions of nodes in the syntactic-semantic tree. As an illustrative example, such an operation can check whether a node belongs to a subtree of another node.

[00051] The coincidence of the pattern defined by the left side of the production rule of the syntactic-semantic structure with the syntactic-semantic structure representing at least part of the sentence of the text in natural language can trigger the right side of the production rule. The right side of the production rule may associate one or more features (related to the lexical, syntactic and (or) semantic properties of the words of the original sentence) with information objects represented by vertices. In one illustrative example, the right side of the interpretation rule may contain a statement linking a token from a natural language text to an aspect of the target entity.

[00052] An identification rule can be used to determine the relationship for a pair of information objects that represent the same entity from the real world. An identification rule is a production rule, the left part of which contains one or more logical expressions related to the nodes of the semantic tree, corresponding to information objects. If a pair of information objects corresponds to a condition defined by logical expressions, information objects are combined into one information object.

[00053] In FIG. 4, at block 410, a computing system implementing method 400 may receive a variety of syntactic-semantic structures representing analyzed sentences from a text in a natural language.

[00054] At block 420, a computing system can detect a variety of syntactic-semantic substructures corresponding to patterns defined on the left side of interpretation rules used to analyze natural language text.

[00055] In blocks 430-450, the computing system can iterate over identified matching instances that can be queued and associated with a priority class (for example, based on a priority class associated with interpretation rules that trigger the corresponding matching instances). At block 430, the computing system can check the assumption that the matching instance queue is empty, if so, the method can go to block 470, otherwise at block 440, the computing system can return the matching highest rated instance from the queue.

[00056] At block 450, the computing system can create one or more logical expressions defined on the right side of the interpretation rules that will switch the matching instance with the highest priority identified at block 340.

[00057] At block 460, the computing system may add a logical expression to the bag of statements associated with the analyzed natural language text. The term “Bag of statements” in this document means a set of logical expressions that do not contradict each other, characterizing the extracted information objects and their properties. Therefore, an attempt to add a new statement to the bag of statements that contradicts at least one statement already in the “bag of statements” will fail.

[00058] After the determination in block 470 of the failure of the attempt to add to the "bag of statements" associated with the analyzed text in natural language, the logical expression obtained in block 450, the method may return to block 430; otherwise, the method may return to block 420.

[00059] At block 480, after it is detected that the queue of matching instances is empty, the method may stop working by declaring the claims bag full. The claims bag can define an RDF graph representing the identified aspect terms. In some embodiments, a computing system may combine some of the identified aspects, for the designation of which various synonymous constructions are used in the source text in a natural language.

[00060] FIG. 5 provides examples of production rules that can be used to identify aspect terms in accordance with one or more aspects of the present invention. Rule 510 may be used to identify an implicit aspect. Rule 520 can be used to identify an implicit aspect. Rule 530 may be used to merge explicit aspects. Rule 540 can be used to identify an implicit aspect that reflects the consumer’s indication that using the product causes back pain. Rule 550 can be used to identify an implicit aspect that reflects the consumer’s indication that part of the product needs to be replaced.

[00061] In FIG. Figure 6 shows some examples of production rules that can be used to determine the tonality of an aspect term in accordance with one or more aspects of the present invention. Rule 610 can be used to identify key related to certain aspects related to passenger cars.

[00062] As noted above in this document, additionally or as an alternative to using production rules, the tonality of an aspect term can be determined by evaluating one or more classifier functions. The scope of the classifier function can be represented by a vector of values of grammatical, syntactic, and (or) semantic features characterizing an aspect term. As the basic component of the syntactic-semantic structure representing the aspect term, the maximum component of the syntactic-semantic structure, the core of which contains the aspect term, can be selected. The context of the identified base component can be represented by parent, child, and sibling components.

[00063] In various illustrative examples, features representing a scope of a classifier function may include:

[00064] a semantic class associated with the identified base component representing an aspect term (eg, HUMAN, BE, EMOTINOS_AND_TH EMOTIONAL_ATTITUDE, INSIDE_PART);

[00065] - a pragmatic class associated with the identified basic component representing an aspect term (for example, DigitalNumbers, Person, Positive, NormalPolarityDependence);

[00066] a language-independent deep position associated with the identified base component representing an aspect term;

[00067] is a language dependent surface position associated with the identified base component representing an aspect term;

[00068] - a syntactic paradigm associated with the identified basic component representing an aspect term, for example, many surface structures associated with certain lexical meanings (for example, parts of speech);

[00069] is a semanthem associated with an aspect term (eg, PolarityPlus, NonPolarityMinus);

[00070] - a capitalization template that defines the use of capital letters in an aspect term (for example, AaAaA, AAAA); and / or

[00071] - the position of the identified basic component in the syntactic-semantic tree (for example, whether the basic component is the root of the collocation or originates from a specific vertex of the semantic tree).

[00072] In some implementations of the invention, feature values may be binarized before being used in the classifier function. In an illustrative example, the binarization procedure includes representing N values of a given attribute in the form of N binary signs. In some embodiments, the resulting binary features can be filtered (for example, using Fisher's exact test or using Shannon's entropy) to select the smallest number of the most statistically significant features.

[00073] Certain parameters of these classifier functions can be adjusted using machine learning methods that use predefined or dynamically generated data sets. A set of supporting data may contain one or more texts in a natural language, in which certain aspects and corresponding tonality are noted. In one illustrative example, a similar set of supporting data can be created or updated using a GUI designed to receive data from a user who selects one or more adjacent words and associates them with the category and tonality of aspect terms.

[00074] In some implementations, classifier functions can be implemented using random forest classifiers. The random forest classifier creates many decision trees during training and generates a class that is a mode of average prediction classes for individual trees.

[00075] In an alternative implementation, classifier functions can implement various machine learning methods, such as linear classifiers, classifiers based on the tree of direct descendants, classifiers based on a random forest, classifiers based on conditional random fields (CRF), classifiers based on latent allocation Dirichlet (LDА), classifiers based on the support vector method (SVM) and (or) classifiers based on neural networks (for example, World2Ves). In one illustrative example, the SVM classifier can create one or more separation hyperplanes in multidimensional space. To determine the values of the hyperplane parameters, a training sample is used so that the obtained hyperplane has a maximum distance from the nearest point of the training data.

[00076] In various illustrative examples, one or more quality metrics can be used to evaluate the quality of the classifier function. These quality metrics can be represented by accuracy (positive predictive value), completeness (sensitivity), or F-measure.

[00077] In some embodiments, aspect-level sentiment analysis performed in accordance with one or more aspects of the present invention may be implemented as an interactive process in which user input is used to provide a user dictionary of various terms related to target entities (for example, goods and services) and for evaluating the results of various operations of the method, as described in more detail below with reference to FIG. 7.

[00078] In FIG. 7 is a flowchart of an example of a method 700 for interactive sentiment analysis at the aspect level in natural language text in accordance with one or more aspects of the present invention. Method 700 and / or each of its individual functions, standard programs, subprograms, or operations can be performed by one or more processors of a computing system (eg, computing system 700 in FIG. 7) that implements this method. In some implementations, method 700 may be implemented in a single processing stream. In an alternative approach, method 700 may be implemented by two or more processing threads, with each thread implementing one or more separate functions, procedures, routines, or operations of this method. In an illustrative example, processing threads implementing method 700 may be synchronized (for example, using semaphores, critical sections, and / or other thread synchronization mechanisms). In addition, the processing threads that implement the method 700 can be performed asynchronously relative to each other. Thus, although FIG. 7 and the corresponding description contain a list of operations for method 700 in a specific order, in various embodiments of the method at least some of the described operations can be performed in parallel and (or) in a random order.

[00079] At block 710, a computing system that implements method 700 may obtain input data defining one or more tokens that can be used in the analyzed natural language text with respect to target entities (eg, consumer goods or services). In one illustrative example, a user dictionary may include a list of terms that can be used in relation to a particular brand or model of a particular device (for example, “iPhone 6”, “iPhone of the sixth generation”, “iPhone of the latest model”, etc.). Additionally or alternatively, user input may define one or more tokens that can be used in the analyzed natural language text to refer to aspects of the target entities. The computing system can use a user dictionary to expand the semantic hierarchy used for syntactic-semantic analysis of natural language text, as described in more detail above with reference to FIG. one.

[00080] In block 720, the computing system may receive natural language text, which may represent, for example, one or more consumer reviews of target entities (eg, consumer goods or services).

[00081] In block 730, the computing device can perform syntactic-semantic analysis of the received text in a natural language. Syntactic-semantic analysis can give at least one language-independent semantic structure for each sentence in a natural language text, as described in more detail below with reference to FIG. 5-15.

[00082] At block 740, the computing system can interpret a variety of syntactic-semantic structures using a set of production rules to determine individual aspects of the target entities, as described in detail below with reference to FIG. 4-5.

[00083] At block 750, the computing system may receive user input that modifies a set of identified aspects. In one illustrative example, user input may delete individual aspect terms that are not of interest to the user, and / or individual incorrectly identified aspect terms. In one illustrative example, user input may add individual aspectual terms that have not been identified but are of interest to the user. In another illustrative example, user-entered data may combine some of the identified aspects, for the purpose of which various synonymous constructions are used in the source text in a natural language. In another illustrative example, user input can change the hierarchical relationship between the identified aspects (for example, the term “battery life” may be associated with the battery, and not with the smartphone itself). In another illustrative example, user input can change the tokens that can be used in the analyzed natural language text to refer to target entities and / or their aspects (for example, the user can define “brakes” as associated with “brake shoes” and “brake cylinders ”). In some embodiments, in response to receiving data from the user, the method may return to block 730, as shown schematically by a dashed line.

[00084] At block 760, the computing system may determine the tonality of the identified aspect terms. In one illustrative example, tonality can be determined by applying a set of production rules to interpret many syntactic-semantic structures obtained as a result of syntactic-semantic analysis of text in a natural language. Additionally or alternatively, tonality can be determined by calculating one or more classifier functions that return the tonality of each identified aspect term, as described in more detail later in this document.

[00085] At block 770, the computing system may receive user input that modifies the identified tonality of aspect terms. In one illustrative example, user input can change the hierarchical relationship between the identified aspects and their keys (for example, user input can associate the term “too expensive” with the price of the target device, and not with the target device itself). In another illustrative example, user input may change a tonality that is incorrectly detected (for example, the word "unpredictable" may reflect a positive sentiment for a movie and at the same time a negative sentiment for a car). In another illustrative example, user input may define an additional aspect term and / or corresponding tonality (for example, user input may associate the term “smooth running” with the characteristic “driving comfort”). In some embodiments, in response to receiving data from the user, the method may return to block 730, as shown schematically by a dashed line.

[00086] At block 780, the computing system may aggregate the extracted information in a form suitable for output in a graphical user interface (GUI) and / or for printing. In an illustrative example, a report created by the system may contain the identifier of the target entity (for example, an electronic device) and one or more hierarchical lists identifying the components, characteristics and (or) functions of the target entity. Each hierarchical list can contain many aspect terms and the corresponding key. In some embodiments of the invention, each hierarchical list may also contain the number of instances (for example, the number of reviews on the network) in which the identified aspect is characterized by the identified aspect term and the identified tonality.

[00087] In response to receiving user input at block 790 with approval of the generated report, the method may stop working at block 795. Otherwise, the computing system may receive user input modifying the user dictionary, and the method may return to block 720.

[00088] In some embodiments, the same user dictionary can be used to analyze natural language text associated with different target entities that have the same characteristics (for example, consumer products or services belonging to the same functional category, such like smartphones, cars, banking, etc.). In one illustrative example, the system can use the same user dictionary to evaluate and compare two different brands / models of a particular consumer product (for example, a smartphone).

[00089] In FIG. 8 is an example natural language text for use in a sentiment analysis method at the aspect level in accordance with one or more aspects of the present invention. The natural language text represents the consumer’s review of the recently purchased vehicle.

[00090] FIG. 9 shows the results of the text markup example shown in FIG. 8 in accordance with one or more aspects of the present invention. An example of markup presents aspects (e.g., general, manageability, convenience, price, etc.), relevant aspect terms and their position in the text, identified tonality (e.g., negative, neutral, positive) and identified types of aspects (e.g. explicit, actual, implicit).

[00091] In FIG. 10 is a flow chart of one illustrative example implementation of a method 200 for performing semantic-syntactic analysis of sentences in natural language 212 in accordance with one or more aspects of the present invention. Method 200 can be applied to one or more syntactic units (eg, sentences) included in a particular text corpus to form a plurality of semantic-syntactic trees corresponding to syntactic units. In various illustrative examples, natural language sentences to be processed by method 200 can be extracted from one or more electronic documents that can be generated by scanning (or by other means of obtaining images of paper documents) and optical character recognition (OCR) to obtain texts corresponding to these documents. Natural language sentences can also be extracted from various other sources, including e-mail messages, texts from social networks, digital content files processed using speech recognition methods, etc.

[00092] In block 214, a computing device that implements this method can conduct a lexical-morphological analysis of sentence 212 to establish the morphological meanings of words included in the sentence. As used herein, the "morphological meaning" of a word means one or more lemmas (i.e., canonical or dictionary forms) corresponding to a word, and a corresponding set of grammatical attribute values that define the grammatical meaning of the word. Such grammatical features may include the lexical category (part of speech) of the word and one or more morphological and grammatical features (for example, case, gender, number, conjugation, etc.). Due to the homonymy and (or) matching grammatical forms corresponding to different lexical and morphological meanings of a particular word, two or more morphological meanings can be established for a given word. A more detailed description of an illustrative example of conducting a lexical-morphological analysis of a sentence is given later in this document with reference to FIG. eleven.

[00093] In block 215, the computing device can perform rough syntax analysis of sentence 212. Rough syntactical analysis may include the use of one or more syntactic models that can be correlated with elements of sentence 212, with the subsequent establishment of surface (ie syntactic) relationships in the framework of Proposition 212 to obtain a graph of generalized components. As used herein, “constituent” means a group of neighboring words of an original sentence that functions as one grammatical entity. A component includes a core in the form of one or more words and may also include one or more child components at lower levels. A child component is a dependent component that can be correlated with one or more parent components.

[00094] At block 216, the computing device can parse sentence 212 accurately to form one or more sentence syntax trees. Among the various syntax trees, based on a specific evaluation function, taking into account the compatibility of the lexical meanings of the words of the original sentence, surface relations, deep relations, etc. one or more of the best syntax trees matching Proposition 212 can be selected.

[00095] At block 217, the computing device can process the syntax trees to form a semantic structure 218 corresponding to Proposition 212. The semantic structure 218 may include many nodes corresponding to semantic classes and may also include many arcs corresponding to semantic relations (for a more detailed description see below. in this document).

[00096] FIG. 11 schematically illustrates an example of the lexical-morphological structure of a sentence in accordance with one or more aspects of the present invention. An example of the lexical-morphological structure 300 may include many pairs of "lexical meaning - grammatical meaning" for an example sentence. As an illustrative example, “And” can be correlated with the lexical meaning “shall” 312 and “will” 314. The grammatical meaning correlated with the lexical meaning 312 is <Verb, GTVerbModal, ZeroType, Present, Nonnegative, Composite II>. The grammatical meaning associated with the lexical meaning 314 is <Verb, GTVerbModal, ZeroType, Present, Nonnegative, Irregular, Composite II>.

[00097] FIG. 12 schematically illustrates language descriptions 210 used, including morphological descriptions 201, lexical descriptions 203, syntactic descriptions 202, and semantic descriptions 204, as well as relationships between them. Among them, morphological descriptions 201, lexical descriptions 203, and syntactic descriptions 202 are language dependent. The set of language descriptions 210 is a model of a particular natural language.

[00098] As an illustrative example, a specific lexical meaning in lexical descriptions 203 may be correlated with one or more surface models of syntactic descriptions 202 corresponding to a given lexical meaning. A specific surface model of syntactic descriptions 202 may be correlated with an in-depth model of semantic descriptions 204.

[00099] FIG. 13 schematically illustrates several examples of morphological descriptions. The components of morphological descriptions 201 may include: definitions of inflection 310, grammar system 320, description of derivation 330 and others. Grammar system 320 includes a set of grammatical categories, such as a part of speech, case, gender, number, person, recurrence, time, type and their meanings (so-called "grammes"), including, for example, an adjective, noun or verb; nominative, accusative or genitive; feminine, masculine or neuter, etc. Corresponding grammars can be used to compose a description of the inflection 310 and a description of the derivation 330.

[000100] Inflection description 310 defines the forms of a given word depending on its grammatical categories (eg, case, gender, number, time, etc.) and broadly includes or describes the various possible forms of the word. Description of word formation 330 determines which new words can be formed from a given word (for example, compound words).

[000101] In accordance with one aspect of the present invention, component models may be used to establish syntactic relationships between elements of the original sentence. A component is a group of neighboring words in a sentence that behave as a whole. The core component is the word; it can also contain child components of lower levels. The child component is a dependent component and can be attached to other components (parent) to build syntactic descriptions 202 of the original sentence.

[000102] In FIG. 14 are examples of syntactic descriptions. The components of syntax descriptions 202 may include, but are not limited to, surface models 410, descriptions of surface positions 420, description of reference and structural control 456, description of control and alignment 440, description of non-wood syntax 450, and analysis rules 460. Syntactic descriptions 202 can be used to build possible syntactic structures of the original sentence in a given natural language, taking into account the free linear word order, non-wood syntactic phenomena (for example, matching , ellipsis, etc.), referential relations and other factors.

[000103] Surface models 410 can be represented as sets of one or more syntactic forms (“synths” 412) to describe possible syntactic sentence structures that are part of syntax description 202. In general, the lexical meaning of a word in a natural language can be associated with surface (syntactic) models 410. The surface model can represent the components that are possible if the lexical meaning acts as a "core". A surface model may include a set of surface positions of child elements, a description of the linear order, and / or diathesis. As used herein, “diathesis” means a definite relationship between surface and depth positions and their semantic roles expressed through depth positions. For example, diathesis can be expressed by the pledge of the verb: if the subject is the agent of the action, the verb is the active pledge, and when the subject is the direction of the action, this is expressed by the passive pledge of the verb.

[000104] A plurality of surface positions 415 of child components and descriptions of their linear order 416 can be used in the component model to describe grammatical values 414 of possible placeholders for these surface positions. Diathesis 417 represents the correspondence between the surface positions 415 and the deep positions 514 (as shown in Fig. 16). Communicative descriptions 480 describe the communicative order in a sentence.

[000105] A description of the linear order (416) can be represented as linear order expressions reflecting the sequence in which various surface positions (415) can occur in the sentence. The expressions of linear order may include the names of variables, the names of surface positions, parentheses, grammes, the operator "or" (or), etc. As an illustrative example, the linear ordering description of the Boys play football simple sentence can be represented as Subject Core Object_Direct, where Subject, Core, Object_Direct are the direct complement names of surface positions 415 corresponding to the word order.

[000106] Communicative descriptions 480 can describe the word order in syntax 412 from the point of view of communicative acts, presented in the form of communicative order expressions that are similar to linear order expressions. Control and reconciliation descriptions 440 may include rules and grammatical constraints on the joining components that are used during parsing.

[000107] Descriptions of non-wood syntax 450 may be created to reflect various linguistic phenomena, such as ellipsis and matching, they are used in transformations of syntactic structures that are created at various stages of analysis in various embodiments of the invention. Descriptions of non-wood syntax 450 may include, but are not limited to, descriptions of ellipsis 452, descriptions of matching 454, and descriptions of referential and structural controls 430.

[000108] Analysis Rules 460 may describe the properties of a particular language and be used as part of semantic analysis. Rules of analysis 460 may include rules for calculating semantems 462 and rules for normalizing 464. Rules for normalizing 464 can be used to describe transformations of semantic structures that may differ in different languages.

[000109] In FIG. 15 provides an example of semantic descriptions. The components of semantic descriptions 204 are language independent and may, among other things, include a semantic hierarchy 510, descriptions of deep positions 520, a system of semantems 530, and pragmatic descriptions 540.

[000110] The core of semantic descriptions can be represented by a semantic hierarchy 510, which can include semantic concepts (semantic entities), also called semantic classes. The latter can be arranged in a hierarchical structure that reflects the parent-child relationship. In general, a child semantic class can inherit one or more properties of its direct parent and other semantic ancestor classes. As an illustrative example, the semantic class SUBSTANCE (Substance) is a child semantic class of the ENTITY class and the parent semantic class for the classes GAS, (Gas), LIQUID (Liquid), METAL (Metal), WOOD_MATERIAL (Wood), etc.

[000111] Each semantic class in the semantic hierarchy 510 may be accompanied by a depth model 512. The depth model 512 of the semantic class may include many depth positions 514, which may reflect the semantic roles of child components in various sentences with objects of a given semantic class as the core of the parent component. The depth model 512 may also include possible semantic classes that act as placeholders for deep positions. Deep positions (514) can express semantic relationships, including, for example, "agent" (agent), "addressee" (destination), "instrument" (instrument), "quantity" (quantity), etc. A child semantic class can inherit and refine the deep model of its immediate parent semantic class.

[000112] The descriptions of the deep positions 520 reflect the semantic roles of the child components in the deep models 512 and can be used to describe the general properties of the deep positions 514. The descriptions of the deep positions 520 may also contain grammatical and semantic restrictions on the placeholders of the deep positions 514. Properties and restrictions associated with with depth positions 514 and their possible placeholders in various languages, can be largely similar and often identical. Thus, the deep positions 514 are language independent.

[000113] The semantem system 530 may be a plurality of semantic categories and semantems that represent semantic category meanings. As an illustrative example, the semantic category "DegreeOfComparison" (Comparison Degree) can be used to describe the degree of comparison of adjectives and include the following semantems: "Positive", "ComparativeHigherDegree" (Comparative Comparison), "SuperlativeHighestDegree" (Superlative Comparison) and others. As another illustrative example, the semantic category "RelationToReferencePoint" (Relation to a point) can be used to describe the order (spatial or temporal in the broad sense of the analyzed words), such as before or after a point or event, and include semantems "Previous" ( Previous) and Subsequent. As another illustrative example, the semantic category "EvaluationObjective" can be used to describe an objective assessment, such as, for example, "Bad", "Good", etc.

[000114] The semantem system 530 may include language-independent semantic attributes that can express not only semantic characteristics, but also stylistic, pragmatic, and communicative characteristics. Some semanthemes can be used to express an atomic meaning, which finds a regular grammatical and (or) lexical expression in natural language. According to their intended purpose and use of the system, semantems can be divided into categories, for example, grammatical semantems 532, lexical semantems 534 and classifying grammatical (differentiating) semantems 536.

[000115] Grammatical semantems 532 can be used to describe the grammatical properties of constituents when transforming a syntax tree into a semantic structure. The lexical semantems 534 can describe specific properties of objects (for example, “being flat” or “being liquid”) and used in descriptions of deep positions 520 as a restriction of placeholders for deep positions (for example, for the verbs “face (with ) "(facing) and" flood "(fill), respectively). Classifying grammatical (differentiating) semantems 536 can express the differential properties of objects within a single semantic class. As an illustrative example, in the semantic class HAIRDRESSER (HAIRDRESSER), the semantema <<RelatedToMen>> (applies to men) is assigned the lexical meaning "barber" in contrast to other lexical values that also belong to this class, for example, "hairdresser", "hairstylist " etc. Using data language-independent semantic properties that can be expressed as elements of a semantic description, including semantic classes, deep positions, and semantems, semantic information can be extracted in accordance with one or more aspects of the present invention.

[000116] The pragmatic descriptions of 540 allow you to assign a specific theme, style or genre to the texts and objects of the semantic hierarchy 510 (for example, “Economic Policy”, “Foreign Policy”, “Law”, “Legislation”, “Trade”, “Finance”, etc. .d.). Pragmatic properties can also be expressed by semantems. As an illustrative example, pragmatic context can be taken into account in semantic analysis.

[000117] In FIG. 16 gives an example of lexical descriptions. Lexical descriptions (203) are the set of lexical meanings 612 of a particular natural language. For each lexical meaning 612, there is a relationship 602 with its language-independent semantic parent in order to indicate the position of any given lexical meaning in the semantic hierarchy 510.

[000118] The lexical meaning 612 in the lexical-semantic hierarchy 510 can be correlated with the surface model 410, which in turn can be correlated through one or more diathesis 417 with the corresponding depth model 512. The lexical value 612 can inherit its parent's semantic class and specify its deep model 512.

[000119] The lexical surface model 410 may include one or more syntactic forms 412. The surface model 410 synth 412 may include one or more surface positions 415, including corresponding descriptions of their linear order 416, one or more grammatical values 414, expressed in in the form of a set of grammatical categories (grammes), one or more semantic restrictions associated with placeholder surface positions, and one or more diathesis 417. Semantic restrictions related Items with a specific placeholder for a surface position can be represented as one or more semantic classes whose objects can fill this surface position.

[000120] In FIG. 17 schematically illustrates examples of data structures that can be used within the framework of one or more methods of the present invention. Referring again to FIG. 10, in block 214, a computing device that implements this method can conduct a lexical-morphological analysis of sentence 212 to construct a lexical-morphological structure 722 according to FIG. 17. The lexical-morphological structure 722 may include many correspondences of lexical and grammatical meanings for each lexical unit (for example, a word) of the original sentence. FIG. 11 schematically illustrates an example of a lexical-morphological structure.

[000121] Returning again to FIG. 10, in block 215, the computing device may conduct a rough parsing of the original sentence 212 to construct a graph of generalized components 732 according to FIG. 17. Rough parsing involves applying one or more possible syntactic models of possible lexical meanings to each element of the set of elements of the lexical-morphological structure 722, in order to establish the set of potential syntactic relations in the initial sentence 212 represented by the graph of generalized components 732.

[000122] The graph of generalized components 732 can be represented by an acyclic graph that includes many nodes corresponding to the generalized components of the original sentence 212 and includes many arcs corresponding to surface (syntactic) positions that can express different types of relations between generalized lexical values. In the framework of this method, many potentially applicable syntactic models can be applied for each element of the set of elements of lexical and morphological structures of the original sentence 212 to form a set of components of the initial sentence 212. Then, within the framework of the method, many possible components of the initial sentence 212 can be considered for constructing a graph of generalized components 732 on based on a set of components. The graph of generalized components 732 at the level of the surface model can reflect many potential connections between the words of the original sentence 212. Since the number of possible syntactic structures can be relatively large, the graph of generalized components 732 can, in general, include redundant information, including a relatively large number of lexical values at certain nodes and (or) surface positions along certain arcs of the graph.

[000123] The graph of generalized components 732 can initially be constructed in the form of a tree, starting from end nodes (leaves) and moving further to the root by adding child components filling the surface positions 415 of the set of parent components so that all lexical units of the original sentence are covered 212.

[000124] In some embodiments, the root of the graph of generalized components 732 is a predicate. In the process described above, a tree can become a graph, as certain lower-level components can be included in one or more upper-level components. Many components that represent certain elements of the lexical-morphological structure can then be generalized to obtain generalized components. Components can be generalized based on their lexical meanings or grammatical meanings 414, for example, on the basis of parts of speech and the relationship between them. In FIG. 18 schematically illustrates an example of a graph of generalized components.

[000125] At block 216, the computing device can parse sentence 212 to form one or more syntax trees 742 of FIG. 17 based on the graph of generalized components 732. For each syntax tree, the computing device can determine the integral estimate based on a priori and calculated estimates. The tree with the best rating can be selected to build the best syntactic structure 746 of the original sentence 212.

[000126] During the construction of the syntax structure 746 based on the selected syntax tree, the computing device can establish one or more non-wood links (for example, by creating an additional link among at least two nodes of the graph). If this process fails, the computing device can select a syntax tree with a conditionally optimal estimate that is closest to the optimal one, and an attempt is made to establish one or more non-wood links in the tree. Finally, as a result of accurate parsing, a syntax structure 746 is created that represents the best syntax structure corresponding to the original sentence 212. In fact, by selecting the best syntax structure 746, the best lexical values 240 for the elements of the original sentence 212 are determined.

[000127] At block 217, the computing device can process the syntax trees to form a semantic structure 218 corresponding to sentence 212. The semantic structure 218 can reflect the semantics transmitted by the original sentence in language-independent terms. The semantic structure 218 can be represented in the form of an acyclic graph (for example, a tree, possibly supplemented by one or more non-wood links (the arc of the graph). The words of the original sentence are represented by nodes with the corresponding language-independent semantic classes of the semantic hierarchy 510. The arcs of the graph represent deep (semantic ) the relationship between the elements of the sentence. The transition to the semantic structure 218 can be carried out using the rules of analysis 460 and involves the correlation of one or more features ( reflecting the lexical, syntactic and (or) semantic properties of the words of the original sentence 212) with each semantic class.

[000128] In FIG. 19 is an example of a syntactic structure of a sentence generated from a graph of generalized components shown in FIG. 18. Node 901 corresponds to the lexical element “life” 906. Using the method of syntactic-semantic analysis described in this document, the computing device can determine that the lexical element “life” 906 represents one form of lexical meaning correlated with with the semantic class "LIFE" 904 and fills the surface position $ Adjunct_Locative 905) in the parent component represented by the Verb: succeed: succeed: TO_SUCCEED control node (907).

[000129] In FIG. 20 shows a semantic structure corresponding to the syntax structure in FIG. 19. With respect to the aforementioned lexical element “life” (906) in FIG. 19, the semantic structure includes the lexical class 1010 and the semantic class 1030 corresponding to those presented in FIG. 19, however, instead of the surface position (905), the semantic structure includes the deep position "Sphere" (scope_of_operation) 1020.

[000130] In accordance with one or more aspects of the present invention, a computing device that implements the methods described herein may index one or more parameters resulting from semantic-syntactic analysis. Thus, the methods of the present invention allow us to consider not only many words in the composition of the original text corpus, but also many lexical meanings of these words, storing and indexing all the syntactic and semantic information obtained in the course of parsing and semantic analysis of each sentence of the original text corpus. Such information may additionally include data obtained during the intermediate stages of the analysis, as well as results of lexical selection, including results obtained during the resolution of ambiguities caused by homonymy and (or) matching grammatical forms corresponding to different lexical and morphological meanings of some words of the original language.

[000131] For each semantic structure, one or more indexes can be created. An index can be represented as a data structure in memory, for example, in the form of a table consisting of several records. Each record can be a correspondence between a particular element of the semantic structure (for example, one word or several words, syntactic relation, morphological, syntactic or semantic property or syntactic or semantic structure) and one or more identifiers (or addresses) of the use of this element of semantic structures in the source text.

[000132] In some embodiments, the implementation of the index may include one or more values of morphological, syntactic, lexical and (or) semantic parameters. These values can be created during a two-stage semantic analysis (for a more detailed description, see this document). The index can be used to perform various natural language processing tasks, including semantic search.

[000133] A computing device that implements this method can extract a wide range of lexical, grammatical, syntactic, pragmatic and (or) semantic characteristics during syntactic-semantic analysis and the creation of semantic structures. In an illustrative example, the system can extract and store certain lexical information, data on the membership of certain lexical units in semantic classes, information on grammatical forms and linear order, information on the use of certain forms, aspects, tonality (for example, positive or negative), deep positions, non-wood connections, semantems, etc.

[000134] A computing device that implements the methods described herein can analyze using one or more of the text analysis methods described in this document and index any one or more parameters of language descriptions, including lexical values, semantic classes, grammes, semantems, and etc. Indexing a semantic class can be used in various natural language processing tasks, including semantic search, classification, clustering, text filtering, etc. Indexing lexical meanings (instead of indexing words) allows you to search not only words and word forms, but also lexical meanings, i.e. words with certain lexical meanings. A computing device that implements the methods of the present invention can also store and index syntactic and semantic structures created by one or more of the text analysis methods described herein to use these structures and (or) indices for semantic search, classification, clustering and filtering of documents .

[000135] In FIG. 21 shows an illustrative example of a computing device (1000) that can execute a set of instructions that cause the computing device to execute any one or more of the methods of the present invention. A computing device may connect to another computing device via a local area network, a corporate network, an extranet, or the Internet. A computing device can operate as a server or client computing device in a client / server network environment, or as a peer computing device in a peer-to-peer (or distributed) network environment. A computing device can be represented by a personal computer (PC), a tablet PC, a television set-top box (STB), a pocket PC (PDA), a cell phone, or any computing device capable of executing a set of commands (sequentially or otherwise) that define the operations that should be performed by this computing device. In addition, while only one computing device is shown, it should be accepted that the term “computing device” may also include any combination of computing devices that separately or jointly execute a set (or several sets) of instructions to perform one or more of the techniques described in this document.

[000136] An example of a computing device (1000) includes a processor (502), main memory (504) (eg, read only memory (ROM) or dynamic random access memory (DRAM)) and a data storage device (518) that communicate with each other on the bus (530).

[000137] The processor (502) may be represented by one or more universal computing devices, for example, a microprocessor, a central processor, etc. In particular, the processor (502) may be a full instruction set microprocessor (CISC), an reduced instruction set microprocessor (RISC), an extra long instruction microprocessor (VLIW), a processor implementing another instruction set, or processors implementing a combination command sets. The processor (502) may also be one or more special-purpose computing devices, for example, a custom integrated circuit (ASIC), field programmable gate array (FPGA), digital signal processor (DSP), network processor, and the like. The processor (502) is configured to execute instructions (526) to perform the operations and functions discussed herein.

[000138] The computing device (1000) may further include a network interface device (522), a visual display device (510), a character input device (512) (eg, a keyboard), and an input device is a touch screen (514).

[000139] The data storage device (518) may comprise a computer-readable storage medium (524) that stores one or more sets of instructions (526) and that implements one or more of the methods or functions of the present invention. The instructions (526) can also be located completely or at least partially in the main memory (504) and / or in the processor (502) while they are being executed in the computing device (1000), with the main memory (504) and the processor (502) also constitute a computer-readable storage medium. Commands (526) can additionally be transmitted or received over the network (516) through a network interface device (522).

[000140] In some embodiments, instructions (526) may include instructions of a method (100) for classifying natural language texts уf based on semantic features and / or a method for calculating classification model parameters. While the computer-readable storage medium (524) shown in the example of FIG. 21 is a single medium, the term “machine-readable medium” should include one medium or several mediums (for example, a centralized or distributed database, and / or corresponding caches and servers) that store one or more sets of instructions. The term “computer-readable storage medium” should also be construed as a term that includes any medium that is capable of storing, encoding or transferring a set of instructions for execution by a machine that causes this machine to execute any one or more of the techniques described in this disclosure. Thus, the term “computer-readable storage medium”, among other things, also refers to solid-state memory and optical and magnetic media.

[000141] The methods, components, and functions described herein may be implemented by discrete hardware components, or they may be integrated into functions of other hardware components, such as ASICS, FPGA, DSP, or similar devices. In addition, methods, components, and functions may be implemented using firmware modules or functional diagrams of hardware devices. The methods, components and functions can also be implemented using any combination of computing tools and software components, or exclusively using software.

[000142] In the above description, numerous details are set forth. However, it will be apparent to those skilled in the art from this description that the present invention can be practiced without these specific details. In some cases, well-known structures and devices are shown in block diagram form, and not in detail, so as not to complicate the description of the present invention.

[000143] Some parts of the description of preferred embodiments are presented in the form of algorithms and a symbolic representation of operations with data bits in computer memory. Such descriptions and representations of algorithms represent the means used by specialists in the field of data processing in order to most effectively transfer the essence of their work to other specialists in this field. In this document and in general, an algorithm is a self-consistent sequence of operations leading to the desired result. Operations require physical manipulations with physical quantities. Usually, although not necessarily, these quantities take the form of electrical or magnetic signals that can be stored, transmitted, combined, compared and subjected to other manipulations. It turned out that, first of all, for normal use, it is convenient to describe these signals in the form of bits, values, elements, symbols, members, numbers, etc.

[000144] However, it should be borne in mind that all of these and similar terms should be associated with the corresponding physical quantities, and that they are simply convenient labels that apply to these quantities. Unless otherwise specifically and explicitly indicated in the following discussion, it should be assumed that throughout the text such terms as “definition”, “calculation”, “calculation”, “calculation”, “receipt”, “establishment”, “change”, etc. .p., relate to the actions and processes of a computing device or similar electronic computing device that works with data and converts data represented in the form of physical (e.g. electronic) quantities in the registers and memory of the computing device into other data similar to immediately presented in the form of physical quantities in the memory or registers of a computing device, or other similar devices for storing, transmitting or displaying information.

[000145] The present invention also relates to a device for performing the operations described herein. Such a device may be specially designed for the required purposes or it may contain a universal computer that is selectively activated or reconfigured using a computer program stored in the computer. Such a computer program may be stored on a computer-readable storage medium, such as, among others, a disk of any kind, including floppy disks, optical disks, compact disks, magneto-optical disks, read-only memory (ROM), random access memory (RAM) ), EPROM, EEPROM, magnetic or optical cards and other types of storage media suitable for storing electronic commands.

[000146] It should be understood that the above description is illustrative and not restrictive. Various other embodiments will become apparent to those skilled in the art after reading and understanding the above description. Therefore, the scope of disclosure should be determined with reference to the appended claims along with the full scope of equivalents to which such claims are entitled.

Claims (31)

1. The method of sentimental analysis of texts in natural language, including: the execution using a computer system of syntactic-semantic analysis of at least part of the text in natural language to obtain many syntactic-semantic characteristics representing part of the text in the natural language; identification in part of the text in the natural language of at least one aspect term representing an aspect associated with the target entity, and at least one syntactic-semantic characteristic associated with the specified at least one aspect term; calculating, using one or more characteristics of the text obtained by syntactic-semantic analysis, the values of the classifier function to determine the tonality associated with the aspect term; creating a report containing one or more hierarchical lists of aspect terms, including identified aspects and tonality of identified aspects; and determining, using a training data sample and a validating data sample, at least one parameter of a classifier function, in which the training data sample includes a learning text in a natural language containing many aspect terms. 2. The method according to p. 1, characterized in that the text in a natural language is a set of consumer reviews of the target entity. 3. The method according to p. 1, characterized in that the target entity is represented by at least a consumer product or service. 4. The method according to p. 1, characterized in that the aspect is represented by at least a characteristic, or function, or component of the target entity. 5. The method according to p. 1, characterized in that the aspect term contains one or more words. 6. The method according to p. 1, characterized in that the tonality associated with the aspect takes one of the following meanings: negative tonality, neutral tonality or positive tonality. 7. The method according to claim 1, characterized in that the classifier function is represented by one of the following options: a linear classifier, a classifier based on a tree of direct descendants, a classifier based on a random forest, a classifier based on conditional random fields (CRF), a classifier based on latent Dirichlet placement (LDA), classifiers based on the support vector method (SVM) or classifier based on neural networks. 8. The method according to p. 1, characterized in that the scope of the classifier function contains one of the following options: the value of the grammatical characteristic characterizing the aspect term, the value of the syntactical characteristic characterizing the aspect term or the value of the semantic characteristic characterizing the aspect term, where the value is obtained by syntactic-semantic analysis. 9. The system of sentimental analysis of texts in natural language, containing: memory; and a processor associated with this memory, configured to: perform syntactic-semantic analysis of at least part of the text in the natural language to obtain many syntactic-semantic structures representing a part of the text in the natural language; identify at least one aspect term related to the aspect related to the target entity and at least one syntactic-semantic characteristic associated with said at least one aspect term in a part of the text in the natural language; calculate, using one or more characteristics of the text obtained by syntactic-semantic analysis, at least one value of the classifier function to determine the tonality associated with the aspect term; create a report containing one or more hierarchical lists of aspect terms, including identified aspects and tonality of identified aspects; and determine, using a training data sample and a validating data sample, at least one parameter of the classifier function, where the training data sample includes a learning text in a natural language containing many aspect terms. 10. The system of claim 9, wherein the aspect relates at least to a characteristic, or function, or component of a target entity. 11. The system of claim 9, wherein the tonality associated with the aspect term can be represented by one of the following meanings: negative tonality, neutral tonality, or positive tonality. 12. The system according to claim 9, characterized in that the classifier function is represented by a classifier based on the support vector method (SVM). 13. The system according to p. 9, characterized in that the processor additionally has the ability to: determine, using a training data set, at least one parameter of a classifier function, the training data set including a learning text in natural language containing a plurality of aspect values. 14. A permanent computer-readable storage medium containing executable instructions that, when the computer system performs sentimental analysis of texts in natural language, induces the computer system: perform syntactic-semantic analysis of at least part of the text in the natural language to obtain many syntactic-semantic structures representing a part of the text in the natural language; identify at least one aspect term related to the aspect related to the target entity and at least one syntactic-semantic characteristic associated with the at least one aspect term in a part of the text in the natural language; calculate, using one or more characteristics of the text obtained by syntactic-semantic analysis, the value of the classifier function to determine the tonality associated with the aspect term; create a report containing one or more hierarchical lists of aspect terms, including identified aspects and tonality of identified aspects; and determine, using a training data sample and a validating data sample, at least one parameter of the classifier function, where the training data sample includes a learning text in a natural language containing many aspect terms.

Images